Customer loyalty in restaurants: myth vs reality

Chapter 1 — Why does a points program with 14,000 sign-ups leave the till flat?

Because the program measures sign-ups and redemptions, while the till moves on FREQUENCY. More than 90% of restaurants already run some rewards scheme (Paytronix, 2025), so having one stopped setting anyone apart:

it is table stakes, like owning a card terminal. Meanwhile average retention across the sector sits near 55% (Restroworks, 2025), and the best quick-service operators barely hold 62% monthly member retention against 57.8% in full service (Paytronix, 2024). With those figures on the table, a dashboard celebrating 14,000 registrations and 900 annual redemptions is celebrating a phone directory, not a repeat-visit engine. The question that actually matters, and that almost no loyalty vendor will put to you, is how many of that member's visits EXIST because of the program and would not have existed without it. A redemption handing contribution margin to a guest who was coming anyway is a discount, not customer loyalty.

Chapter 2 — A redemption is not an incremental visit, and confusing them costs the most

Here sits the trade's paradox: the program reporting the most redemptions is usually the one giving away the most margin to its most loyal base, the people who were never going to leave. Diego F. Parra insists at Masterestaurant on splitting two dashboard columns before anyone discusses budget: the member's visits before enrolling and their visits after, measured across the same window of months. If the gap fails to reach 15%, that program costs money. And the sector hints at where the real muscle lives: email marketing returns US$36 for every US$1 (Stripo, 2025), with a 43.6% open rate in restaurants (Stripo, 2025). The cheap channel exists. What usually goes missing is the REASON sent through it. Transactional loyalty carries a low, declining acquisition cost per member, yet it is structurally incapable of bringing in anyone new: it only works with whoever already walked through the door.

Chapter 3 — Transactional loyalty versus a content engine: two different cost curves

Content behaves the opposite way, variable in cost and sensitive to format, and it remains the only entrance to the funnel. Some 78% of restaurants already use Instagram (Restroworks, 2025), where engagement reaches 2.2% against Facebook's 0.22% (Restroworks, 2025), a tenfold gap, and that is where a random Tuesday's repeat visit gets decided today. With a 7.1% conversion rate on Google Ads in the food category (WordStream, 2025), working out what you can afford to pay for a new guest stops being intuition and becomes arithmetic. My position is firm: buying transactional loyalty alone optimizes the exit of the funnel while the entrance dries up. The same program produces opposite results depending on the size of the till, which is why I always break it out. Below US$500,000 a year, no loyalty license pays for itself: there the asset is an SMS list, with 98% open rates (Constant Contact, 2024) and 45% response against email's 6% (Omnisend, 2025).

Chapter 4 — What changes across annual revenue bands

Between US$500,000 and US$1 million the first profitable case appears, provided email already returns its US$36 per dollar (Stripo, 2025). Above US$1 million the operation can carry a part-time content editor feeding Instagram, where engagement runs at 2.2% (Restroworks, 2025). Above US$5 million the problem shifts from cost to attribution across locations. Above US$10 million, with 55% retention as the benchmark (Restroworks, 2025), one point of retention is worth more than the entire marketing line. Past US$5 million a year, the celebrity-chef restaurant or the large-format themed venue plays a different game, and its customer loyalty carries costs that rarely reach the budget. The chef's personal brand generates free reach, true, but it also sets a permanent floor of audiovisual production: influencer marketing returns somewhere between US$5.78 (Socially Powerful, 2025) and US$7.65 per dollar invested at a 2.55% average conversion (iQFluence, 2026), and that multiplier switches off the moment cadence drops.

Chapter 5 — The high end: celebrity chefs, big formats and costs nobody budgets

Add the drag of gift cards, whose unredeemed value hovers around 6% (Capital One Shopping, 2026): cash already collected that stays a booked liability until somebody eats it. In this bracket the defensible asset is not the points program, it is the owned audience, and that one you either rent or build. Take the scenario all the way through before deciding. Month one: you save the license and lose the redemptions, so average ticket RISES because the discount to your loyal base disappears. Month three: the list is still alive, email holds its 43.6% open rate (Stripo, 2025), and base visits do not move, because they never depended on the point. Month six: if you published nothing with a reason behind it, frequency starts sliding and you discover the program was propping up a reminder, however poor. That is the lesson: what you switched off was not the discount, it was the contact.

Chapter 6 — What would happen if you switched the program off tomorrow and kept only content?

With SMS at 98% open rates (Constant Contact, 2024) and 18% click-through (Tabular, 2025), replacing the reminder costs pennies. Replacing the reason costs editorial work.

What it costs to bring them in, what they leave each time, how many times they come back. There is no fourth. If your customer loyalty program moves none of the three within twelve months, it is operating expense dressed up as strategy and it deserves shutting down without ceremony. For years I argued the opposite, that the registration alone counted as an asset; I was wrong, and that three-location owner's dashboard, eighteen months flat, was the argument I could not answer. Sector data backs the correction: average retention of 55% (Restroworks, 2025) even though more than 90% of restaurants already run a program (Paytronix, 2025). When nearly everyone owns the tool and the indicator refuses to climb, the tool was never the cause.

Chapter 7 — The three figures that judge a dish also judge loyalty

Diego F. Parra and the Masterestaurant method treat loyalty as frequency engineering, not as a prize catalog. Transactional loyalty pays fast with a low ceiling; content pays slowly with a high one, and that lag explains why so many owners abandon the second just before it starts earning. Order your spending by horizon rather than enthusiasm: first the direct contact channel, cheapest and most immediate, with email at 43.6% open rates (Stripo, 2025) and SMS at 98% (Constant Contact, 2024); then short-form video on Instagram, where 2.2% engagement (Restroworks, 2025) multiplies Facebook tenfold; and only then paid media, using its 7.1% conversion rate in the food category (WordStream, 2025) as your control floor. This week do one single thing: cross your member list against the last six months of visits and calculate what share came back more than twice. The program measures REDEMPTIONS; the content engine measures REPEAT VISITS.

Chapter 8 — The five differences that decide margin

They are different things, and only the second one shows up in the income statement. A redemption can easily be a guest who was already coming, taking contribution margin without adding an incremental visit, and that reading error is the costliest one on customer loyalty dashboards. Customer acquisition cost behaves in opposite ways across the two models. Transactional loyalty carries a low, declining CAC per enrolled guest and ZERO ability to bring in new people; content carries a variable CAC, sensitive to format, and it is the only entrance to the sales funnel. With a 7,1% conversion benchmark on Google Ads for food (WordStream, 2025), deciding what to pay for a new guest stops being intuition. The payback horizon differs by a factor of three. A program needs critical mass, so it takes nine to eighteen months to show a clean effect; an SMS campaign to your own list moves cash the same weekend, with a 45% reply rate against 6% for email (Omnisend, 2025).

Chapter 9 — The five differences that decide margin — in practice

For an owner with tight cash, that time difference is not a nuance, it is the entire decision. Value leakage has an opposite nature in each model. In transactional loyalty the leak is accounting-grade and silent: roughly 6% of gift card value is never redeemed (Capital One Shopping, 2026), money that entered as cash and stayed as a permanent liability. In content the leak is attention, and it warns you: email opens slide below the 43,6% sector benchmark (Stripo, 2025) weeks before traffic falls. The competitive moat differs too. Your neighbor copies a points program in thirty days using the same vendor; a content library with its own voice, real cases and a community, no. That is why the Masterestaurant framework places the program as data INFRASTRUCTURE and content as the strategic asset, never the other way around.